OpenAI has terminated three researchers, including Mikita Balesni, Tomek Korbak, and Jasmine Wang. The former employees claim they were fired for communicating with external AI safety organizations, a task they assert was part of their professional duties.
Balesni, who focused on AI alignment, stated that his interactions with external safety groups were approved by management. Korbak noted that his dismissal was linked to his engagement with METR, a non-profit organization that evaluates AI models and safety risks.
In a public letter, the researchers warned that these sudden dismissals could foster a culture of fear, discouraging staff from openly discussing AI risks. They recommended that OpenAI involve external safety auditors and maintain transparent communication with the broader safety ecosystem.
OpenAI management stated in an internal memo that they support the researchers' recommendations and value open discourse. However, the company officially maintained that the terminations followed an investigation into the violation of policies regarding the handling of confidential information.
